Genetic manipulation of candidate phyla radiation bacteria provides functional insights into microbial dark matter

Abstract

The study of bacteria has yielded fundamental insights into cellular biology and physiology, biotechnological advances and many therapeutics. Yet due to a lack of suitable tools, the significant portion of bacterial diversity held within the candidate phyla radiation (CPR) remains inaccessible to such pursuits. Here we show that CPR bacteria belonging to the phylum Saccharibacteria exhibit natural competence. We exploit this property to develop methods for their genetic manipulation, including the insertion of heterologous sequences and the construction of targeted gene deletions. Imaging of fluorescent protein-labeled Saccharibacteria provides high spatiotemporal resolution of phenomena accompanying epibiotic growth and a transposon insertion sequencing genome-wide screen reveals the contribution of enigmatic Saccharibacterial genes to growth on their Actinobacteria hosts. Finally, we leverage metagenomic data to provide cutting-edge protein structure-based bioinformatic resources that support the strain Southlakia epibionticum and its corresponding host, Actinomyces israelii , as a model system for unlocking the molecular underpinnings of the epibiotic lifestyle.

Figure 1.

Phylogenetic placement and genome sequencing of newly isolated Saccharibacteria strains S. epibionticum ML1 (Se) and N. lyticus ML1 (Nl). (A) Maximum growth (fold change) achieved by Se and Nl during co-culture with compatible host species A. israelii and Propionibacterium propionicum respectively, and population change (growth or death) detected at equivalent timepoints with an incompatible host. (B) Phylogeny of CPR and other bacteria based on concatenated ribosomal proteins, with the placement of the Saccharibacteria phylum indicated. Figure adapted from Castelle et al. (C) Phylogeny constructed using 49 core, universal genes indicating placement of Se and N. l (blue text) within Saccharibacteria. Family names (as designated by the Genome Taxonomy Database) and groups previously designated by McLean et al. (G1, etc.) are indicated for each clade. HOT, human oral taxon. (D, E) Overview of the genome sequences of S. epibionticum ML1 and N. lyticus ML1.

Figure 2.

Harnessing natural transformation to generate mutant Saccharibacteria. (A) Schematic depicting the intergenic neutral site (blue, NS1) targeted for insertion of a hygomycin resistance cassette (yellow) in the Se genome and the linear DNA fragment employed in transformation experiments. Primer binding sites used for genotyping are indicated (sites 1–3). (B) Overview of the Se transformation protocol. After incubation with linear DNA, Se + Ai co-cultures are enlarged concomitant with hygromycin addition and serially passage with addition of naïve host at each dilution to promote Se growth (grey box). Clonal transformed Se populations were obtained by plating to isolate single colonies of Ai with accompanying Se cells, followed by growth in liquid culture, with additional Ai, to promote Se population expansion. (C) PCR-based genotyping of Se clones obtained following transformation according to the protocol show in (B) in the presence (right) or absence (left) of selection with hygromycin during the expansion and passaging steps. Binding sites for primers targeting NS1 (1, 2) and hph (3) are shown in (A). Positive control primers (Se) target a locus distant from NS1. (D) Growth (red) and percent of Se transformed (grey) over the course of transformation protocol depicted in (B), in the presence (squares) or absence (circles) of selection with hygromycin. (E) Luminescence production from Se–Ai co-cultures (left) or co-culture filtrates (right) in which Se contains a nanoluciferase expression cassette inserted at NS1 (shown at bottom). (F) Fluorescence and phase contrast micrographs of Se–Ai co-cultures in which Se carries an mcherry (top) or sfgfp (bottom) expression cassette inserted at NS1. See also Figure S2. Data in (E) represent mean ± s.d. Asterisks indicate statistically significant differences (unpaired two-tailed student’s t-test; *p<0.05).

Figure 3.

Fluorescent protein expression and quantitative microscopy enable tracking of the S. epibionticum lifecycle. (A) and (D) Snapshots captured at the indicated time points from timelapse fluorescence and phase contrast microscopy of GFP-expressing Se grown in co-culture with Ai. Arrows indicate Se cells exhibiting productive (pink, purple) and non-productive (blue) interactions with Ai cells. White outlines in the fluorescent channel indicate depict an Ai cell affected by Se infection (Ai 1). (B) and (E) Omnipose-generated segmentation of Se and Ai cells depicted in (A) and (D), at the start (left) and end (right) of the 20 or 22 hr growth period. (C) and (F) Growth of individual Ai cells as impacted by productive (light grey) or non-productive Se cells (black, dark grey). Colors correspond to cell masks shown in (B) and (E). For the full time course captured in (A) and (D), see Videos S1 and S2, respectively. For additional examples of tracked Se–Ai growth, see Videos S3-S11.

Figure 4.

Identification of genes important for fitness of S. epibionticum during co-culture with Ai identified by Tn-seq. (A) Overview of normalized transposon insertion frequency across the Se genome detected in input DNA used for mutagenesis (dark grey), and from samples collected immediately following transformation (T0, dark blue) and subsequent outgrowth time points (T1-T3, shades of blue). Genes encoding proteins belonging to CPR-enriched protein families (light green) and those found to be significantly important for fitness across all time points and by two different metrics (dark green). The location of the arginine deiminase system (ADS) genes and two loci containing T4P genes (T4P_1, T4P₂) are indicated in the outer circle. (B) Schematic depicting T4P₁ and flanking genes (separated by vertical lines) and the relative fitness contribution of each gene as determined by the TRANSIT resampling algorithm (adjusted p-values) and ALDEx2 (delta scores). Asterisks indicated genes found to be significant by both metrics across all time points. Gene annotations were derived from Foldseek queries using AF models generated for each gene product (see methods). (C) Se population levels detected in Se–Ai cocultures following transformation with constructs designed to replace the indicated genes with hph. Candidate essential genes tested were considered significant for Se fitness by TRANSIT and ALDEx2 at a minimum of 2 of 3 time points (see Table S1). (D, E) Total Se population (D) and proportion transformed (E) following transformation with an unmarked cassette targeted to NS1 in the indicated strains of Se. See also Figures S3 and S4, Table S1. Data in (C-E) represent mean ± s.d. Asterisks indicate statistically significant differences (C, one-way ANOVA followed by Dunnett’s compared to no DNA control; E, unpaired two-tailed student’s t-test; *p<0.05, ns, not significant).

Figure 5.

Inclusion of extensive metagenomic data in MSAs enables proteome-wide AF modeling of Se protein structures. (A) Histograms depicting MSA depths obtained for Se proteins using HHblits. (B) Maximum depths obtained for Se protein MSAs that initially contained <500 sequences. Additional sequences were sourced from metagenomic sequence databases and incorporated into MSAs using Jackhmmer or Phmmer (see Methods). (C) Comparison of the AF confidence metric (pLDDT) determined using Hhblits or Jackhmmer/Phmmer (Metagenome)-generated MSAs for Se proteins with initially shallow MSAs (<500). Se proteins shown in (D) and (E) are highlighted in blue. (D and E) Example Se protein structure models and associated predicted alignment matrices obtained using shallow (right) or metagenomic sequence-improved (left) MSAs. Se proteins models (blue) are aligned to models from top FoldSeek (FS) hits (light grey, A0A2H0BDY7 (D), A0A8B1YPH4, metagenome and A0A3D0YBM7, shallow, (E)), when available. The annotation in (D) derives from the best FS hit; in (E), the best FS hit is an uncharacterized protein. The function for this protein was assigned using DPAM. Some structures are trimmed to highlight the alignment. See also Figure S5, Table S2.